• Journal of Internet Computing and Services
• /
• v.21 no.1
• /
• pp.127-135
• /
• 2020

As the importance of the use and analysis of big data is emerging, there is a growing interest in natural language processing techniques for unstructured data such as news articles and comments. Particularly, as the collection of big data becomes possible, data mining techniques capable of pre-processing and analyzing data are emerging. In this case study with a telecom company, we propose a methodology how to formalize unstructured data using text mining. The domain is determined as equipment failure and the data is about 2.2 million equipment check ledger data. Data on equipment failures by 800,000 per year is accumulated in the equipment check ledger. The equipment check ledger coexist with both formal and unstructured data. Although formal data can be easily used for analysis, unstructured data is difficult to be used immediately for analysis. However, in unstructured data, there is a high possibility that important information. Because it can be contained that is not written in a formal. Therefore, in this study, we study to develop digital transformation method for unstructured data in equipment check ledger.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2023.11a
• /
• pp.185-187
• /
• 2023

고장 진단은 IoT 장비의 안전성과 효율성을 유지하는데 필요한 기술 중 하나이다. 따라서, 본 연구는 IoT 센서 데이터를 기반한 고장 진단 알고리즘을 개발하는데 목적이 있다. 본 연구는 알고리즘의 효율성을 개선하기 위해 기술통계량을 기반하여 데이터 차원을 축소하였으며, 이를 바탕으로 고장 진단 알고리즘의 정확도 및 연산시간을 개선하였다. 본 연구는 다양한 후보 알고리즘을 활용하여 고장진단을 수행하였으며, 정확도를 기반으로 가장 우수한 알고리즘을 선정하였다. 연구 결과, Isolation Forest 알고리즘이 가장 뛰어난 분류 결과를 나타내었다. 본 연구결과를 통해 IoT 센서의 안전성과 신뢰성을 향상시키는 데 도움을 줄 수 있다.

• ETRI Journal
• /
• v.14 no.4
• /
• pp.97-105
• /
• 1992

본 고에서는 TDX-1A 시스팀의 고장데이타를 신뢰도 측면에서 분석하여 운용중 발생하는 부품의 평가고장률을 도출하였다. 또한 도출된 평가고장률을 MIL-HDBK-217방법에 의해 구한 예측고장률과 비교 분석함으로써, TDX-1A 시스팀 신뢰도 예측시 적용했던 MIL-HDBK-217 예측방법이 실제 운용데이터보다 높게 나타나고 있음을 제시하였다. IC 부품에 대해 외국의 부품고장률 계산식을 살펴보고, 그에 따른 고장률 계산결과를 살펴봄으로써, 운용데이터를 이용한 운용환경에 적합한 부품고장률 예측식의 설정 연구가 필요함을 아울러 제시하였다.

• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.876-877
• /
• 2015

전동기의 고장 진단을 함에 있어 데이터의 수집은 필수적으로 이루어 져야 하며 데이터를 수집하는 과정에서 데이터의 손상이나 손실을 없애기 위해 어플리케이션을 만들었다. 제작한 어플리케이션은 모터의 전압과 전류, 소음과 진동을 실시간으로 측정 할 수 있으며, 고장 여부를 판단 할 수 있도록 파일로 저장 가능하게 만들었고, 저장하는 과정에서 데이터가 손실 되지 않도록 버퍼를 사용하였다. 더 효율적으로 측정이 가능하도록 하여 전동기의 고장 진단에 있어 확실한 데이터를 얻을 수 있도록 연구하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.9B
• /
• pp.1295-1304
• /
• 2010

This paper proposes a fault-tolerant data aggregation which provides energy efficient and reliable data collection in wireless sensor networks. The traditional aggregation scheme does not provide the countermeasure to packet loss or the countermeasure scheme requires a large amount of energy. The proposed scheme applies caching and re-transmission based on the track topology to the adaptive timeout scheduling. The proposed scheme uses a single-path routing based on the traditional tree topology at normal, which reduces the dissipated energy in sensor nodes without any countermeasure against packet loss. The proposed scheme, however, retransmits the lost packet using track topology under event occurrences in order to fulfill more accurate data aggregation. Extensive simulation work under various workloads has revealed that the proposed scheme decrease by 8% in terms of the dissipated energy and enhances data accuracy 41% when the potential of event occurrence exists as compared with TAG data aggregation. And the proposed scheme decrease by 53% in terms of the dissipated energy and shows a similar performance in data accuracy when the potential of event occurrence exists as compared with PERLA data aggregation.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.11a
• /
• pp.88-91
• /
• 2020

발전 시설은 연간 무중단으로 운영되어야 하고, 고장이 발생하면 손해가 막대하기 때문에 발전 시설 장비에는 수십만 개의 센서 데이터가 설치되어 있다. 본 논문에서는 효율적인 센서 데이터의 수집과 시설 모니터링 및 고장 예측 등을 위한 빅데이터 스트리밍 질의 처리 시스템을 설계 및 구현하였다. 또한 실시간 데이터 수집의 효율적인 관리를 위해 인코딩 방식을 설계하였으며, 데이터 전송 성능을 측정하여 문자열로 데이터를 전송하는 것보다 평균 12%, 최대 32% 데이터 처리 성능이 향상됨을 보였다. 또한, 스트리밍 데이터에 대한 윈도우 질의 처리 성능을 측정하여 약 0.97초의 평균 집계 질의 처리 시간이 소요됨을 확인하였다. 향후에는 고장 감지를 위한 인공지능 추론 모델을 제안하는 빅데이터 스트리밍 질의 처리 시스템에 적용할 예정이다.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.12 no.2
• /
• pp.29-37
• /
• 1998

This paper describes the fault diagnosis based on real-time data of the inverter system for brush less DC motor drive. After identifying all the fault types in the inverter system, a preliminary typical analysis of fault types has been classified into the key fault symptoms. The predicted fault performances are then substantiated by using ACSL(Advanced Continuous Simulation Language), and the simulated results are composed of knowledge-base. The real-time measured data from the inverter system are compared with the simulated knowledge-base through the inference engine of expert system, which have been used to diagnose the fault causes. If some faults may occur in the inverter system, this system will be stopped. And then the expertise of elimination and remedial strategies about the fault causes, will be supplied rapidly to operator who doesn't know well about the inverter drive system.system.

• The KIPS Transactions:PartD
• /
• v.10D no.5
• /
• pp.829-836
• /
• 2003

Software Reliability Growth Model that have been studied variously. But measurement of correct parameter of this model is not easy. Specially, estimation of correct model about failure data must be establish and estimation of parameter can consist exactly. To get correct testing, we calculate the normal score and describe the normal probability plot. Use the normal probability plot, we estimate the distribution for failure data. In this paper, we estimate the software reliability growth model for through the normal probability plot. In this research, we applies software reliability growth model through distribution characteristics of failure data. If we see plot, we determine the software reliability growth model, we can make sure superior in model's performance estimation.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.11 no.2
• /
• pp.203-208
• /
• 2016

This paper presents a fault detection and data processing algorithm for acoustic sensor systems using the multiple sensor algorithm that has originally developed for the wireless sensor nodes. The multiple sensor algorithm can increase the reliability of the sensor systems by utilizing and comparing the measurements of the multiple sensors. In the acoustic sensor system, the equivalent sound level($L_{eq}$) is used to detect the faulty sensor. The experiment was conducted to demonstrate the feasibility of the multiple acoustic sensor algorithm, and the results show that the algorithm can detect the faulty sensor and validate the data.

• Journal of KIISE
• /
• v.44 no.5
• /
• pp.460-468
• /
• 2017

In a real-time cloud environment, the data analysis framework plays a pivotal role. Spark Streaming meets most real-time requirements among existing frameworks. However, the framework does not meet the second scale real-time fault recovery requirement. Spark Streaming fault recovery time increases in proportion to the transformation history length called lineage. This is because it recovers the last state data based on the cumulative lineage recorded during normal operation. Therefore, fault recovery time is not bounded within a limited time. In addition, it is impossible to achieve a second-scale fault recovery time because it costs tens of seconds to read initial state data from fault-tolerant storage. In this paper, we propose two techniques to solve the problems mentioned above. We apply the proposed techniques to Spark Streaming 1.6.2. Experimental results show that the fault recovery time is bounded and the average fault recovery time is reduced by up to 41.57%.